RDFVector: An Efficient and Scalable Schema for Semantic Web Knowledge Bases

As the semantic web grows in popularity and enters the mainstream of computer technology, RDF(Resource Description Framework) datasets are becoming larger and more complex. As datasets grow larger and more datasets are linked together, scalability becomes more important. As more complex ontologies are developed, there is growing need for efficient queries that handle inference. In areas such as research, it is vital to be able to perform queries that retrieve not just facts but also inferred knowledge and uncertain information. Currently, scalability and performance issues limit the semantic web from reaching its full potential. The primary goal of RDFVector is to improve the performance of queries against large RDF datasets, including queries that involve inferred knowledge and probabilistic reasoning. RDFVector is a new persistent data model using bit vectors that is highly efficient. This design allows us to store inferred knowledge with no penalty. Queries become simpler and more efficient. 2 Proposed Approach Figure 1 provides a high level diagram of our architecture. Double boxes indicate tables. The process flow is clearly indicated by the directional arrows. The solid arrows indicate actions that can change data, and the dashed arrows query data. The add and delete actions operate against the triples table. The triples table calls the inference engine to add or delete inferred triples. After determining triples to add or delete, the inference engine calls the triples table to perform the action. This cycle demonstrates the recursive nature of our inference solution. One can see from this diagram that three tables provide the core RDFKB schema: the triples table, PSTable and POTable. There is a single one way arrow connecting them showing that all updates to the bit vectors are triggered by the triples table. The query engine supports the user's queries and utilizes the bit vector tables to execute the queries efficiently